We showed previously that actin flow at the IS influences integrin activation via a mechanical coupling mechanism (23, 38). We, therefore, wondered whether the converse is also true—i.e., whether integrin engagement influences actin flow, and if so, whether the β1 and β2 integrins VLA-4 and LFA-1 behave similarly. To test this, we conducted careful side-by-side analysis of T cells stimulated with VCAM-1 (a ligand for VLA-4) and ICAM-1 (a ligand for LFA-1). Jurkat T cells stably expressing GFP–actin were imaged by spinning disk confocal microscopy while spreading on glass coverslips coated with anti-CD3 alone, or together with either ICAM-1 or VCAM-1. Actin flow rates were determined as a function of cell radius based on kymographic analysis, as described in Section “Materials and Methods” and detailed in Ref. (40). As we reported previously (30), Jurkat cells spreading on anti-CD3 alone exhibited fastest actin flow in the peripheral lamellipodial region; flow rates decreased with inward movement until a central region of low actin density was reached (Figures 1A,B; Video S1 in Supplementary Material). Addition of VCAM-1 to the anti-CD3 coated stimulatory surface dramatically slowed actin flow across the entire IS. By contrast, addition of ICAM-1 did not slow actin flow; flow rates were indistinguishable from cells stimulated with anti-CD3 alone (Figure 1B; Videos S2 and S3 in Supplementary Material). Finally, stimulation with all three ligands led to slowing similar to that seen with anti-CD3 together with VCAM-1, demonstrating that the slowing effect observed upon engagement of β1 integrin dominates cell behavior. As a simpler way of expressing these phenotypes, we considered actin flow rates within the lamellipodial region alone (defined as the outer 20% of the IS radius; shaded region in Figure 1B). Within this region, Jurkat T cells spreading on anti-CD3 alone or anti-CD3 + ICAM-1 exhibited actin flow rates of 85–95 nm/s, whereas co-engagement of VCAM-1 (alone or together with ICAM-1) slowed flow to about 15 nm/s (Figure 1E).

Since there are multiple abnormalities in the transformed Jurkat T cell line, we asked if the differential effects of VLA-4 and LFA-1 engagement are recapitulated in primary human T cells. CD4⁺ T cells were purified and expanded from human peripheral blood and transduced with GFP–Lifeact. When stimulated with anti-CD3, these primary human CD4⁺ T lymphoblasts exhibited actin flow rates that were slightly faster than those in Jurkat T cells (Figure 1F; Videos S4–S7 in Supplementary Material). As in Jurkat cells, addition of VCAM-1 to the stimulatory surfaces retarded actin flow. Interestingly, however, primary T cells differed from Jurkat T cells in that costimulation with ICAM-1 retarded actin flow to a similar extent as VCAM-1. A modest additional decrease was observed in cells responding to anti-CD3 together with both integrin ligands.

The differences we observe between Jurkat T cells and primary human CD4⁺ T lymphoblasts can be explained, at least in part, by differences in cell surface integrin expression. Flow cytometric analysis revealed that in comparison with primary T cells, the Jurkat T cells express relatively low levels of LFA-1 (Figure 1C). Indeed, the mean fluorescence intensity of Jurkat T cells was approximately fivefold lower than that of primary T cells. Note that since Jurkat T cells have ~3-fold more surface area than primary human T cells, LFA-1 surface density is approximately 15-fold lower than that of primary T cells. On the other hand, Jurkat T cells constitutively express very high levels of VLA-4, even when compared with primary human CD4⁺ T cells after 7 days of stimulation, a time when their VLA-4 expression is maximal (Figure 1D and data not shown). Moreover, the primary T cell population is more heterogeneous; some cells express high levels of VLA-4 comparable to Jurkat T cells, while others are nearly negative. This probably accounts for the bimodal distribution of actin flow rates observed when these cells engage anti-CD3 and VCAM-1 (Figure 1F). In support of this view, one population of cells (presumably the high VLA-4 expressors) exhibits actin retardation similar to that seen in Jurkat T cells. In any case, these results show that in primary human T lymphoblasts, engagement of either β1 or β2 integrins dramatically slows centripetal actin flow.

Engaged integrins could retard the flow of the F-actin network through mechanical coupling (i.e., by exerting drag), by signaling to actin-regulatory proteins, or both. To assess the contribution of mechanical coupling, we asked if the mobility of integrin ligands influences actin flow rates. Studies parallel to those shown in Figure 1 were performed, except that instead of immobilizing anti-CD3, VCAM-1, and ICAM-1 on glass coverslips, these ligands were used to functionalize supported lipid bilayers. Ligands were attached to the lipid head groups so that they were as mobile as the lipids themselves, and FRAP analysis was performed to verify free ligand mobility (data not shown). Figure 2A shows actin flow rates in Jurkat T cells stimulated using mobile ligands, plotted beside rates in cells responding to immobile ligands (from Figure 1E) to facilitate comparison. Flow rates in Jurkat cells responding to immobile and mobile anti-CD3 alone were comparable. As observed with immobile ligands, addition of VCAM-1, but not ICAM-1, slowed actin flow in Jurkat cells. However, on mobile surfaces, VCAM-1 only slowed actin flow to 60 nm/s, as opposed to 15 nm/s on immobile glass surfaces. Primary T cells behaved similarly to changes in ligand mobility (Figure 2B, plotted with data from Figure 1F). Here, costimulation with either ICAM-1 or VCAM-1 retarded actin flow, and the magnitude of retardation was diminished if integrin ligands were presented on mobile surfaces. The observation that slowing of actin flow is maximal when T cells interact with low mobility integrin ligands supports a mechanical model in which ligand-bound integrins engage actin filaments and exert drag forces that oppose centripetal flow. By contrast, the finding that actin flow is slowed significantly even when ligands are freely mobile points to the involvement of more complex biology, since viscous drag should be very low in this experimental condition. Additional analysis will be required to determine the basis of this effect, but it seems likely that biochemical mechanisms involving regulation of actin polymerization are also involved.

On the surface of bona fide APCs, MHC Class II molecules are uniformly distributed and freely mobile, while integrin ligands cluster in punctate regions where their mobility is constrained by interaction with the cortical cytoskeleton (38). To model this situation, we generated mixed-mobility surfaces bearing patterns of immobilized integrin ligands surrounded by planar bilayers coated with mobile anti-CD3. Three different patterns varying in spot size, spacing, and geometry were created (Figures 3A–C), and the effect of each pattern on T cell actin flow was tested. To facilitate analysis, these studies were performed using Jurkat T cells, which are larger and exhibit more uniform actin dynamics than primary cells, and using VCAM-1 as the immobilized integrin ligand. Actin flow rates were assessed within the lamellipodial region, choosing areas that maximally overlay immobile VCAM-1 spots (on pattern, yellow dashed lines in Figures 3A–C) or bypass them completely (off pattern, blue dashed lines in Figures 3A–C). Figure 3D shows actin flow rates for each region, in comparison with the response to a uniform mobile bilayer coated with anti-CD3 alone (indicated by a red dashed line). Actin flow was dramatically slowed in T cells interacting with patterns A and B (hexagonal arrays with relatively close spacing). Interestingly, slowing was observed whether measurements were made on or off the pattern, indicating that slowing effects are distributed over the actin network, away from sites of integrin engagement. Actin flow was also strongly slowed in T cells responding to pattern C (a square array with more open spacing), but only for on-pattern measurements. More modest retardation was observed off pattern using this array, where the conduit between integrin spots is significantly wider than in patterns A and B.

To further probe the behavior of the actin network, we analyzed T cells responding to mixed-mobility surfaces where VCAM-1 is organized in bars (Figure 3E). As shown in Figures 3E,F, actin flowed freely in regions where the free edge of the lamellipodium fell outside the pattern (outside, see white line in Figure 3E), was slowed dramatically along the bars (on pattern), and showed intermediate flow rates between bars (off pattern). Measurements made at various points parallel to the bars revealed that flow rates increase with distance from the nearest site of integrin engagement (Figure 3G). Taken together, these results show that the effects of localized integrin engagement are distributed across the actin network, impacting actin dynamics many microns away from the site of integrin engagement. This finding further emphasizes the mechanical nature of integrin effects on the overall actin network.

To explore the relationship between integrin engagement, actin flow, and TCR signaling, we labeled spreading T cells with antibodies that detect early tyrosine phosphorylation events in the TCR signaling cascade. As shown in Figures 4A–C, Jurkat T cells spreading on anti-CD3 together with increasing doses of VCAM-1 exhibit dose-dependent diminution of overall tyrosine phosphorylation, as well as diminished phosphorylation of ZAP-70 at the activating residue Y319. In previous work, Nguyen et al. showed that VCAM-1 engagement augmented TCR signaling as measured by Ca²⁺ flux and NF-AT activation (39). Since Nguyen et al. used a lower concentration of anti-CD3 than that used here, we considered the possibility that integrin-dependent enhancement of signaling can only be observed under conditions of sub-optimal TCR triggering. To test this, we titered down the dose of anti-CD3 until minimal signaling was observed. In our experimental system, this occurred at 0.4 µg/ml OKT3 (Figures 4D–F). Nonetheless, at this dose of anti-CD3 (and at all other doses tested), engagement of VLA-4 resulted in reduced overall tyrosine phosphorylation, as well as reduced phosphorylation of both ZAP-70 and PLCγ1 at activating sites.

As shown in Figures 5G-I, integrin engagement also diminished tyrosine phosphorylation in primary human CD4⁺ T lymphoblasts. While the most profound effects were observed after costimulation with VCAM-1, significant diminution in signaling was also observed after costimulation with ICAM-1. Thus, in these cells, engagement of either VLA-4 or LFA-1 diminishes signaling. This is especially important because adhesion at the IS is dominated by LFA-1–ICAM-1 interactions, making this the physiologically relevant receptor–ligand pair.

Our finding that integrin engagement diminishes tyrosine phosphorylation at the IS was surprising, given that integrins are well known to costimulate T cell activation. All of the cells we analyzed were well spread, indicating that they had, in fact, responded to a TCR stimulus. To confirm this, we tested the effect of integrin engagement on Ca²⁺ signaling, the earliest detectable event after TCR triggering. As shown in Figure 4J and Figure S1 in Supplementary Material, stimulation of Jurkat T cells in the presence or absence of VCAM-1 gave identical Ca²⁺ responses, and this result was obtained for two different doses of OKT3. The finding that early Ca²⁺ responses are unaffected by integrin ligands is consistent with the known requirement for initial TCR signaling as a prerequisite for actin polymerization and integrin activation. Taken together, our results are consistent with a model in which the earliest TCR signaling events set actin flow into motion at the IS, thereby fostering additional force-dependent tyrosine phosphorylation, while also activating an integrin-dependent mechanism that governs this mechanotransduction.

Based on our signaling studies, proteins involved in coupling integrins to actin flow should be important modulators of T cell activation. Linkage of integrins to the actin cytoskeleton is best studied in mesenchymal cells, which form well-defined focal adhesions and related adhesive structures. Such structures are not typically observed in T cells, which lack actin stress fibers and generate highly dynamic adhesions associated with low traction forces (33, 34, 41). Canonical focal adhesion proteins, such as vinculin, talin, and paxillin are enriched at the IS (42–44), and in protrusive invadopodium- or podosome-like structures (26, 45–47), but how adhesion complexes are organized at the IS has not been carefully examined. We, therefore, used surfaces bearing patterned integrin ligands and a uniform field of immobilized anti-CD3 to impose structure and ask if focal adhesion proteins are recruited to T cell adhesive contacts. To verify that ICAM-1 patterns correspond to sites of integrin engagement, we first labeled primary human CD4⁺ T lymphoblasts with Kim 127 and m24, antibodies that specifically detect the extended (intermediate affinity) and extended-open (high affinity) conformations of LFA-1, respectively (48, 49). We have previously used these antibodies to analyze T cells spreading on a uniform field of ICAM-1, and shown that the extended conformation of LFA-1 (detected with Kim 127) is enriched in lamellipodial protrusions, while the extended-open conformation (detected with m24) is enriched in an inner ring corresponding to the lamellar region, where actin-dependent forces induce conformational change (23). The distribution of LFA-1 activation intermediates was similar for T cells responding to patterned ICAM-1 (Figure 5A). Consistent with its higher ligand-binding affinity, the extended-open conformation colocalized strongly with spots of ICAM-1.

We next labeled cells with antibodies to talin, vinculin, and paxillin, to ask if these proteins are recruited to T cell adhesive contacts. As shown in Figures 5B,C, Jurkat T cells showed co-localization of talin, vinculin, and paxillin with VCAM-1 patterns. Similar results were obtained with primary human T cells interacting with either VCAM-1 or ICAM-1 patterns (Figures 5D,E), although talin labeling was not as robust in these cells. Labeling with phalloidin confirmed that even when large adhesive contacts are imposed by substrate patterning, these contacts are not associated with actin stress fibers.

Having observed recruitment of canonical focal adhesion proteins to patterns of engaged integrins, we more closely examined cells interacting with a uniform field of integrin ligand to ask if focal adhesion-like complexes could be detected. As shown in Figure 6, primary human T cells spreading on anti-CD3 together with either VCAM-1 or ICAM-1 do sometimes exhibit well-organized adhesive structures containing vinculin, talin, and paxillin. Such structures begin appearing once spreading is maximal (approximately 15 min after addition of cells) and become more prominent over the next 10–20 min (data not shown). In many cases, these structures are elongated and oriented perpendicular to the lamellipodium (inset in Figure 6B). This distribution is consistent with traction mapping studies showing that at late time points of spreading, T cells exert radially arrayed forces on the interacting substrate (33, 35).

Talin and vinculin serve as “clutch molecules” that couple engaged integrins to the dynamic actin network (19). Thus, we hypothesized that loss of these molecules would relieve integrin-dependent retardation of actin flow and de-repress TCR signaling. To test this idea, we generated Jurkat T cell lines expressing GFP–actin together with shRNAs targeting talin or vinculin. As controls, we generated a cell line expressing the empty suppression vector, and another expressing control shRNA targeting α-actinin (a focal adhesion protein that is expected to lack clutch activity). Western blot analysis verified that talin and vinculin levels were reduced to less than 10% of control levels (Figure 7A). Only about 50% reduction of α-actinin levels could be achieved, probably reflecting the fact that the shRNA used targets only α-actinin 4, while the available antibody detects both α-actinin 1 and 4. Each of the suppressed cell lines was analyzed by immunofluorescence microscopy using VCAM-1 patterns (Figures 7C–F). Note that only planes near the coverslip are shown, and the vinculin antibody shows significant background staining. As shown in Figures 7D,E, suppression of either talin or vinculin led to loss of both talin and vinculin from VCAM-1 patterns. This result is consistent with evidence from non-hematopoietic cells that these two proteins interact with activated integrins and actin via a positive feedback loop (50, 51). Recruitment of vinculin to talin has been shown to involve mechanical force, which unmasks vinculin binding sites within the talin stalk region (52, 53). To ask if this holds true in T cells, cells were treated with the Rho-kinase inhibitor Y-27632 together with the actin stabilizing agent Jasplakinolide, a cocktail that arrests actin flow and relieves force on engaged integrins (23, 29, 30). As shown in Figure 7F, this results in loss of both talin and vinculin from VCAM-1 patterns. Thus, we conclude that talin and vinculin are recruited cooperatively to sites of integrin engagement, via a process that depends on ongoing actin flow.

Since suppression of either talin or vinculin resulted in loss of both clutch proteins from sites of integrin engagement, we asked whether this relieves integrin-dependent retardation of actin flow. As expected, all cell lines exhibited comparable actin flow rates when stimulated with OKT3 alone, and retardation of actin flow was observed in untransduced cells and empty vector controls stimulated with OKT3 together with VCAM-1 (Figure 7B; Videos S8–S11 in Supplementary Material). Importantly, suppression of either vinculin or talin significantly reversed the ability of VCAM-1 engagement to retard actin flow. This reversal of integrin-dependent retardation was more pronounced for talin-suppressed cells, which showed actin flow rates approaching those of control cells stimulated with OKT3 alone. This is consistent with models in which talin is the primary linker of integrins to actin, with vinculin reinforcing this connection. Reversal of integrin-dependent retardation was specific to cells deficient for talin or vinculin; cells expressing the α-actinin shRNA behaved like the untransduced and empty vector controls. Taken together, these results show that talin and vinculin work together to couple engaged integrins to the dynamic T cell actin network.

Finally, we asked whether in addition to reversing the effects of integrin engagement on actin flow, suppression of talin or vinculin would also relieve inhibition of TCR-dependent tyrosine phosphorylation. As shown in Figure 8A, suppression of either vinculin or talin partially reversed the loss of total tyrosine phosphorylation induced by VCAM-1 engagement. As with actin flow, the effect was greatest for talin-suppressed cells, which showed phospho-tyrosine intensity approaching that of control cells stimulated with anti-CD3 alone. Similar results were obtained for phospho-ZAP70 (Figure 8B), although the reversal was not as complete. One unexpected finding from this study is that vinculin suppression augmented signaling in cells stimulated with anti-CD3 alone. The basis for this effect is unclear, but a similar result was also obtained for ZAP-70. Since vinculin suppression did not accelerate actin flow rates in T cells stimulated on anti-CD3 alone (Figure 7B), this increase in TCR signaling cannot be attributed to changes in actin flow. Thus, vinculin may impact early TCR signaling events in an actin-independent fashion, as well as via modulation of integrin-dependent actin dynamics. Additional analysis will be required to address this possibility.

Analysis of the relationship between tyrosine phosphorylation levels and actin flow rates for multiple experimental manipulations revealed a simple, linear correlation (R² = 0.63). Linear regression analysis showed that the point representing vinculin-suppressed cells stimulated with OKT3 alone fell outside the 95% confidence interval (data not shown). When this outlying data point was omitted from the analysis, the remaining points correlated strongly (R² = 0.90) (Figure 8C). The simplest explanation for this relationship is that actin flow is a key element in determining the magnitude of early signaling events downstream of TCR engagement, and that integrins tune TCR signaling by modulating actin flow.

Unless otherwise noted, reagents were purchased from Sigma-Aldrich. Y-27632 and Blebbistatin were from EMD Millipore. Jasplakinolide, Alexa Fluor 568-phalloidin, and Alexa Fluor 488-Phalloidin were from Thermo Fisher, and CF405-Phalloidin was from Biotium. Unless otherwise specified, antibody labeling kits (Alexa Fluor 488, 568, 594, and 647) were from Thermo Fisher. Streptavidin was from Jackson ImmunoResearch Laboratories.

The following antibodies were used for immunofluorescence microscopy: rabbit anti-vinculin (Cat. #73412) was from Abcam, and rabbit anti-paxillin (Cat. # sc-5574) was from Santa Cruz Biotechnology. Talin was detected with mouse monoclonal 8D4 (Sigma), after direct conjugation to CF488 using Mix-n-Stain antibody labeling kit (Biotium). Phospho-tyrosine was detected with PY20 (Santa Cruz; #sc-508), directly conjugated to AF-488. Phosphorylated ZAP70 (Y319) and phospholipase C gamma 1 (PLCγ1, Y783) were detected with rabbit polyclonal antibodies from Santa Cruz (#sc-12946-R) and Cell Signaling (#2821), respectively. Secondary antibodies (donkey anti-Rabbit AF-488 and AF-568) were from Thermo Fisher.

For immunoblotting, total α-actinin (1 and 4) was detected using Sigma anti-α-actinin 1 (Cat. # HPA006035). Talin was detected using 8D4, and vinculin was detected with the mouse monoclonal hVin-1 (Sigma), or the Abcam rabbit polyclonal. For flow cytometry, anti-human CD11a–FITC (BD Pharmingen Cat. #55383) was used to label LFA-1, while anti-human CD29-APC (BD Pharmingen # 559883), was used to label VLA-4. Detection of LFA-1 activation intermediates was performed using monoclonal antibodies Kim127 (hybridoma from ATCC), and m24 (Abcam). Kim127 detects both the extended and extended-open conformations of LFA-1, whereas m24 specifically binds to the high-affinity extended-open conformation (48, 49, 76–79).

Artificial stimulatory surfaces were generated using the following reagents: goat anti-human IgG (Fc specific) was from Fitzgerald. Mouse anti-human CD3 (OKT3) was from BioXCell, biotinylated OKT3 was from eBioscience, and human ICAM-1-Fc and VCAM-1-Fc chimeras were from R&D Systems. His-tagged human VCAM-1 Fc was from Sino Biological Inc., and His-tagged human ICAM-1 Fc was purified from cells as previously described (80). For patterning, ICAM-1 and VCAM-1 were labeled with AF-647 using a microscale labeling kit (Invitrogen).

The human Jurkat T lymphoma cell line stably expressing GFP–actin (81) was grown in RPMI 1640 containing Penicillin/Streptomycin, Glutamax (all from Gibco Thermo Fisher) and 5% FBS/5% newborn calf serum (Atlanta Biological) at 37°C in 5% CO₂. Primary human peripheral blood CD4⁺ T cells were obtained without donor identifiers from the Univ. of Pennsylvania’s Human Immunology Core under an Institutional Review Board approved protocol. Lymphoblasts were generated by activation with human T-Activator CD3/CD28 magnetic beads (Dynabeads, Life Technologies) in RPMI supplemented with 10% FBS, 1% GlutaMAX, Penicillin/Streptomycin and 50 U/ml of human rIL-2 (obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health; human rIL-2 from M. Gately, Hoffmann-LaRoche). T lymphoblasts were cultured at 37°C in 5% CO₂. Beads were magnetically removed on day 7 after initial stimulation, and cells were then cultured for an additional day in the presence of 10 U/ml of IL-2.

TRC shRNA targeting constructs in pLKO.1-puro were obtained from the Univ. of Pennsylvania High Throughput Screening Core. For each target protein, at least five shRNAs were tested, and the clones yielding the best suppression were selected for generation of stably suppressed cell lines. Optimal targeting constructs were as follows: shTalin: CCCAGAGTATTAACGCTCCAA; shVinculin: CGGTTGGTACTGCTAATAAAT; shα-Actinin 4: GCCACACTATCGGACATCAAA. Recombinant lentivirus was generated in 293T cells co-transfected using the calcium phosphate method with each DNA of interest (in pLX301 for expression or pLKO.1-puro for suppression), together with psPAX2 and pMD2.G. Medium was exchanged after 18 h, and supernatant harvested 24 h later was immediately used to transduce T cells.

Primary human CD4⁺ T cells expressing GFP–Lifeact (82) were generated by lentiviral transduction as detailed previously (23). Briefly, cells were induced by spin infection with lentivirus on day 1 after activation with human T-Activator beads. Lentivirus-containing media was replaced with T cell culture media, and after 3 days, puromycin was added to a concentration of 2 µg/ml. Activator beads were removed at day 7 after activation, and cells were used at days 8–9. GFP–actin Jurkat T cells stably suppressed for talin, vinculin, and α-actinin were generated similarly, but without activation. Suppression was monitored by Western blotting throughout the study.

Cells were lysed in 1% NP-40 (IGEPAL), 50 mM Tris, pH 8.0, 150 mM NaCl, 5 mM KCl, and 5 mM MgCl₂, 10 mM NaF, 1 mM NA₃VO₄, and complete protease inhibitor cocktail (Roche) for 20 min at 4°C. Lysates were diluted in NuPage sample buffer and separated on NuPage 4–12% Gradient gels (Thermo Fisher). After transferring to nitrocellulose membranes and blocking with 0.5× blocking buffer (Licor Biosciences), proteins were probed with the indicated primary antibodies followed by appropriate secondary antibodies, and then imaged using an Odyssey fluorescence scanner (Li-Cor Biosciences). Images were captured within the linear range for each probe, and sample intensities were normalized to GAPDH staining.

T cells were washed and suspended in FACS buffer (PBS, 5% FBS, 0.02% NaN₃, 1 mM EDTA) containing appropriate antibodies. After washing, cells were analyzed on an LSRII flow cytometer (BD Biosciences). Analysis was carried out using FlowJo (Tree Star).

Stimulatory surfaces were prepared as detailed in Ref. (40). For ligands immobilized on glass surfaces, coverslips (for fixed cell studies) or Sticky-Slide VI 0.4 chambers (Ibidi, for live cell studies) were coated with 10 µg/ml OKT3 for 2 h at 37°C or overnight at 4°C, washed, and incubated with 2 µg/ml ICAM-1 Fc and/or VCAM-1 Fc for 2 h at 37°C. For studies using supported planar lipid bilayers, lipids (DOPC, DSPE-PEG2000-biotin, and DOGS-NTA nickel salt; Avanti Polar Lipids, Inc.) were reconstituted in chloroform at 89.9:0.1:10 mol%, dried and then hydrated in PBS. Lipids were sonicated, and then passed through a 50-nm pore membrane using a mini-extruder (Avanti Polar Lipids, Inc.) to generate small unilamellar vesicles. 25 × 75-mm glass coverslips were cleaned using Piranha solution (3:1 ratio of sulfuric acid and 30% hydrogen peroxide) or Nochromix (Godax Laboratories Inc.), washed with distilled water, air dried, and applied to Sticky-Slide VI 0.4 chambers. Small unilamellar vesicles were added to the chambers to cover the exposed glass surface for 15 min, after which chambers were rinsed with PBS, and incubated with His-tagged ICAM-1 or VCAM-1 (2 µg/ml), followed by 1 µg/ml streptavidin for 15 min. After rinsing, chambers were then incubated with 10 µg/ml OKT3-biotin for 15 min. Chambers were rinsed and left in L-15 medium supplemented with 2 mg/ml d-glucose (imaging medium). Lipid bilayers were used for imaging studies on the same day.

To prepare mixed-mobility stimulatory surfaces, PDMS stamps with patterns were cast on fabricated silicon wafers. Goat anti-human IgG antibodies (50 µg/ml) were deposited onto the stamps, washed with distilled water, and dried. Patterns were then stamped onto Nochromix-cleaned coverslips for 1 min, after which the coverslips were dried and adhered to Sticky-Slide VI 0.4 chambers. Chambers were then incubated with small unilamellar vesicles containing biotinylated lipids (DOPC:DSPE-PEG2000-biotin 99.9:0.1) as described above. After thorough washing with PBS, chambers were incubated with ICAM-1 Fc or VCAM-1 Fc (each labeled with AF-647), streptavidin and OKT3-biotin, maintained in imaging medium, and used the same day.

For fixed cell studies on patterned surfaces, Nochromix-cleaned coverslips were stamped directly with 20 µg/ml VCAM-1 or ICAM-1 labeled with AF-647 as described above. The stamped coverslips were then coated with 10 µg/ml OKT3 for 2 h at 37°C or overnight at 4°C, washed once with PBS, and twice with imaging medium prior to use.

Fixed and living cells were imaged on a Perkin Elmer Ultraview spinning disk confocal system, using a 63x Plan Apo, 1.4N.A. objective mounted on a Axiovert 200M microscope platform (Carl Zeiss) equipped with an Orca ER CCD camera (Hamamatsu). Images were acquired and analyzed using Volocity v. 6.3 imaging software (Perkin Elmer).

For fixed cell imaging, T cells were harvested, resuspended in imaging medium, and pipetted onto stimulatory coverslips. Cells were allowed to spread for the indicated times at 37°C, after which they were washed in PBS and fixed in 3% paraformaldehyde/PBS. After quenching with 50 mM NH₄Cl, cells were permeabilized with 0.1% Tx-100/PBS and blocked with PBS/0.01% saponin/0.25% fish skin gelatin. Cells were labeled with primary antibodies followed by secondary antibodies, as indicated. Primary antibodies against pZAP70, pTyr, and pPLCγ1 were incubated overnight at 4°C, otherwise antibody labeling was for 1 h at RT. Because both the stimulatory antibody OKT3 and anti-talin antibody 8D4 are mouse monoclonals, anti-talin was directly conjugated to CF488 to avoid crossreactivity with the secondary antibody. After antibody labeling, F-actin was visualized with fluorescent phalloidin, nuclei were stained with either 0.1 µg/ml DAPI (Sigma), or 5 mM Draq5 (eBioscience), and coverslips were mounted onto slides with Mowiol 4-88 (Sigma). Labeling with conformation-specific integrin antibodies was performed as described previously (23). Kim127 was conjugated to AF-594, and LFA-1 was directly conjugated to AF-488. Because the epitope recognized by m24 is destroyed by fixation, cells were labeled with this antibody for 5 min prior to fixation. Cells were then fixed and processed as described above.

Live-cell imaging was conducted as previously described (30, 40). Ibidi μ-Slide VI chambers coated with stimulatory ligands were filled with imaging medium and equilibrated at 37°C. Cells were resuspended at 1 × 10⁶/ml in imaging medium and pipetted into each well. Imaging began 3–5 min after cell addition to allow the cells to contact and spread on the chamber bottoms. Three image planes were captured through focus in 0.25 µm increments and collected every 1–2 s for 4 min. Emission discrimination filters were used for multicolor imaging. During the next 30 min, a series of 4–5 movies, consisting of 5–8 cells per field, were collected for 4 min each. Fields to be analyzed were selected at random, and analysis was restricted to well-spread cells in which actin behavior was clearly visible. Actin flow was measured at 9–15 sites per cell, in 20–60 total cells accumulated from at least three independent experiments.

Kymographic analysis was performed using Volocity v. 6.3, as detailed in Ref. (30, 40). Briefly, kymographs were generated by drawing a ray from the center of IS to its periphery. Actin flow rates were calculated based on the angle of deflection from the vertical direction. IS radius values from individual cells were normalized to 1, and instantaneous velocities were plotted as a function of the IS radius and grouped into 10 equally spaced bins. Flow rates were calculated in the lamellipodium region, defined as the normalized cell radius in the range of 0.8–1. For actin flow analysis on patterned surfaces, kymographs were generated by choosing areas that maximally intersected the immobile spots (on pattern), bypassed them completely (off pattern), or lay outside the patterned region (outside).

To measure tyrosine phosphorylation, the relative amounts of pTyr, pZAP-70, and pPLCγ1 were determined by labeling with relevant phospho-specific antibodies, and then measuring the fluorescence intensity at the IS. Briefly, T cells were allowed to settle and spread onto stimulatory surfaces for 5 min to activate signaling, and fixed for 5 min before labeling. Sixteen fields were imaged near the plane of the coverslip, each containing multiple cells. In each case, a z-series containing 3-planes spaced 0.25 µm apart was captured. The integrated signal intensity from the rendered stack was measured using Volocity 6.3 by manually drawing a ROI around each cell. Only intact cells with a well-spread morphology were analyzed. Data represent mean ± SD from 50 to 60 cells per condition.

Jurkat T cells were loaded with 2 µM Fura-2 acetoxymethyl ester (Thermo-Fisher) in a buffer solution containing 145 mM NaCl, 4.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, 10 mM HEPES, 2 mM glutamine, and 2% fetal bovine serum for 15 min at room temperature. For imaging, cells were washed and resuspended in buffer solution. Stimulatory surfaces were prepared as described above. Ca²⁺ responses were imaged using a Zeiss Plan-Neofluor 40× oil objective (N.A. 1.3) on a Zeiss Axiovert 200 M inverted microscope equipped with an MS-2000 automatic stage (Applied Scientific Instruments) and a 37°C environmental chamber. A Lambda DG-4 Xenon light source (Sutter) was used for rapid transition between 340 and 380 nm excitation. Images were collected for 30 min at 1 to 2-s intervals using Slidebook 6.0 software (Intelligent Imaging Innovation). Analysis of Ca²⁺ data was done using ImageJ software (NIH, Bethesda, MD, USA). Ca²⁺ signal was analyzed by plotting the emission ratio of 340/380-nm excitation for each cell. Individual plots were shifted to correct for different touch down times and an average plot was created for each condition.

Statistical analysis was performed using Microsoft Excel and Prism software. Unless otherwise indicated, statistical significance was calculated using Student’s t-test for unpaired samples, and data are presented as mean ± SD.